A data platform that pairs a data lake's inexpensive file storage with the table structure and transactions of a data warehouse.
A Data Lakehouse combines what a data lake and a data warehouse each do well in one platform.
Traditionally, data lakes stored large amounts of unstructured data cheaply but had poor query performance, while data warehouses performed well but were rigid and expensive. The lakehouse gets both by keeping the data in open files on cloud object storage and running warehouse-style tables and transactions directly over those files.
This architecture suits AI workloads, because data scientists can run large-scale machine learning on the unified data layer without first building complex ETL pipelines to move the data.
See also Machine Learning, Data Mobilization, Vector Database and Data Sovereignty.